CINXE.COM

<!DOCTYPE html> <html lang="en"> <head> <meta charset="utf-8"> <meta name="viewport" content="width=device-width, initial-scale=1.0"> <meta name="color-scheme" content="light dark"> <title>PEP 557 – Data Classes | peps.python.org</title> <link rel="shortcut icon" href="../_static/py.png"> <link rel="canonical" href="https://peps.python.org/pep-0557/"> <link rel="stylesheet" href="../_static/style.css" type="text/css"> <link rel="stylesheet" href="../_static/mq.css" type="text/css"> <link rel="stylesheet" href="../_static/pygments.css" type="text/css" media="(prefers-color-scheme: light)" id="pyg-light"> <link rel="stylesheet" href="../_static/pygments_dark.css" type="text/css" media="(prefers-color-scheme: dark)" id="pyg-dark"> <link rel="alternate" type="application/rss+xml" title="Latest PEPs" href="https://peps.python.org/peps.rss"> <meta property="og:title" content='PEP 557 – Data Classes | peps.python.org'> <meta property="og:description" content="This PEP describes an addition to the standard library called Data Classes. Although they use a very different mechanism, Data Classes can be thought of as “mutable namedtuples with defaults”. Because Data Classes use normal class definition syntax, y..."> <meta property="og:type" content="website"> <meta property="og:url" content="https://peps.python.org/pep-0557/"> <meta property="og:site_name" content="Python Enhancement Proposals (PEPs)"> <meta property="og:image" content="https://peps.python.org/_static/og-image.png"> <meta property="og:image:alt" content="Python PEPs"> <meta property="og:image:width" content="200"> <meta property="og:image:height" content="200"> <meta name="description" content="This PEP describes an addition to the standard library called Data Classes. Although they use a very different mechanism, Data Classes can be thought of as “mutable namedtuples with defaults”. Because Data Classes use normal class definition syntax, y..."> <meta name="theme-color" content="#3776ab"> </head> <body> <svg xmlns="http://www.w3.org/2000/svg" style="display: none;"> <symbol id="svg-sun-half" viewBox="0 0 24 24" pointer-events="all"> <title>Following system colour scheme</title> <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"> <circle cx="12" cy="12" r="9"></circle> <path d="M12 3v18m0-12l4.65-4.65M12 14.3l7.37-7.37M12 19.6l8.85-8.85"></path> </svg> </symbol> <symbol id="svg-moon" viewBox="0 0 24 24" pointer-events="all"> <title>Selected dark colour scheme</title> <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"> <path stroke="none" d="M0 0h24v24H0z" fill="none"></path> <path d="M12 3c.132 0 .263 0 .393 0a7.5 7.5 0 0 0 7.92 12.446a9 9 0 1 1 -8.313 -12.454z"></path> </svg> </symbol> <symbol id="svg-sun" viewBox="0 0 24 24" pointer-events="all"> <title>Selected light colour scheme</title> <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"> <circle cx="12" cy="12" r="5"></circle> <line x1="12" y1="1" x2="12" y2="3"></line> <line x1="12" y1="21" x2="12" y2="23"></line> <line x1="4.22" y1="4.22" x2="5.64" y2="5.64"></line> <line x1="18.36" y1="18.36" x2="19.78" y2="19.78"></line> <line x1="1" y1="12" x2="3" y2="12"></line> <line x1="21" y1="12" x2="23" y2="12"></line> <line x1="4.22" y1="19.78" x2="5.64" y2="18.36"></line> <line x1="18.36" y1="5.64" x2="19.78" y2="4.22"></line> </svg> </symbol> </svg> <script> document.documentElement.dataset.colour_scheme = localStorage.getItem("colour_scheme") || "auto" </script> <section id="pep-page-section"> <header> <h1>Python Enhancement Proposals</h1> <ul class="breadcrumbs"> <li><a href="https://www.python.org/" title="The Python Programming Language">Python</a> » </li> <li><a href="../pep-0000/">PEP Index</a> » </li> <li>PEP 557</li> </ul> <button id="colour-scheme-cycler" onClick="setColourScheme(nextColourScheme())"> <svg aria-hidden="true" class="colour-scheme-icon-when-auto"><use href="#svg-sun-half"></use></svg> <svg aria-hidden="true" class="colour-scheme-icon-when-dark"><use href="#svg-moon"></use></svg> <svg aria-hidden="true" class="colour-scheme-icon-when-light"><use href="#svg-sun"></use></svg> Toggle light / dark / auto colour theme </button> </header> <article> <section id="pep-content"> <h1 class="page-title">PEP 557 – Data Classes</h1> <dl class="rfc2822 field-list simple"> <dt class="field-odd">Author:</dt> <dd class="field-odd">Eric V. Smith <eric at trueblade.com></dd> <dt class="field-even">Status:</dt> <dd class="field-even"><abbr title="Accepted and implementation complete, or no longer active">Final</abbr></dd> <dt class="field-odd">Type:</dt> <dd class="field-odd"><abbr title="Normative PEP with a new feature for Python, implementation change for CPython or interoperability standard for the ecosystem">Standards Track</abbr></dd> <dt class="field-even">Created:</dt> <dd class="field-even">02-Jun-2017</dd> <dt class="field-odd">Python-Version:</dt> <dd class="field-odd">3.7</dd> <dt class="field-even">Post-History:</dt> <dd class="field-even">08-Sep-2017, 25-Nov-2017, 30-Nov-2017, 01-Dec-2017, 02-Dec-2017, 06-Jan-2018, 04-Mar-2018</dd> <dt class="field-odd">Resolution:</dt> <dd class="field-odd"><a class="reference external" href="https://mail.python.org/pipermail/python-dev/2017-December/151034.html">Python-Dev message</a></dd> </dl> <hr class="docutils" /> <section id="contents"> <details><summary>Table of Contents</summary><ul class="simple"> <li><a class="reference internal" href="#notice-for-reviewers">Notice for Reviewers</a></li> <li><a class="reference internal" href="#abstract">Abstract</a></li> <li><a class="reference internal" href="#rationale">Rationale</a></li> <li><a class="reference internal" href="#specification">Specification</a><ul> <li><a class="reference internal" href="#field-objects"><code class="docutils literal notranslate">Field</code> objects</a></li> <li><a class="reference internal" href="#post-init-processing">post-init processing</a></li> <li><a class="reference internal" href="#class-variables">Class variables</a></li> <li><a class="reference internal" href="#init-only-variables">Init-only variables</a></li> <li><a class="reference internal" href="#frozen-instances">Frozen instances</a></li> <li><a class="reference internal" href="#inheritance">Inheritance</a></li> <li><a class="reference internal" href="#default-factory-functions">Default factory functions</a></li> <li><a class="reference internal" href="#mutable-default-values">Mutable default values</a></li> <li><a class="reference internal" href="#module-level-helper-functions">Module level helper functions</a></li> </ul> </li> <li><a class="reference internal" href="#discussion">Discussion</a><ul> <li><a class="reference internal" href="#python-ideas-discussion">python-ideas discussion</a></li> <li><a class="reference internal" href="#support-for-automatically-setting-slots">Support for automatically setting <code class="docutils literal notranslate">__slots__</code>?</a></li> <li><a class="reference internal" href="#why-not-just-use-namedtuple">Why not just use namedtuple?</a></li> <li><a class="reference internal" href="#why-not-just-use-typing-namedtuple">Why not just use typing.NamedTuple?</a></li> <li><a class="reference internal" href="#why-not-just-use-attrs">Why not just use attrs?</a></li> <li><a class="reference internal" href="#post-init-parameters">post-init parameters</a></li> <li><a class="reference internal" href="#asdict-and-astuple-function-names">asdict and astuple function names</a></li> </ul> </li> <li><a class="reference internal" href="#rejected-ideas">Rejected ideas</a><ul> <li><a class="reference internal" href="#copying-init-false-fields-after-new-object-creation-in-replace">Copying <code class="docutils literal notranslate">init=False</code> fields after new object creation in replace()</a></li> <li><a class="reference internal" href="#automatically-support-mutable-default-values">Automatically support mutable default values</a></li> </ul> </li> <li><a class="reference internal" href="#examples">Examples</a><ul> <li><a class="reference internal" href="#custom-init-method">Custom __init__ method</a></li> <li><a class="reference internal" href="#a-complicated-example">A complicated example</a></li> </ul> </li> <li><a class="reference internal" href="#acknowledgements">Acknowledgements</a></li> <li><a class="reference internal" href="#references">References</a></li> <li><a class="reference internal" href="#copyright">Copyright</a></li> </ul> </details></section> <section id="notice-for-reviewers"> <h2><a class="toc-backref" href="#notice-for-reviewers" role="doc-backlink">Notice for Reviewers</a></h2> This PEP and the initial implementation were drafted in a separate repo: <a class="reference external" href="https://github.com/ericvsmith/dataclasses">https://github.com/ericvsmith/dataclasses</a>. Before commenting in a public forum please at least read the <a class="reference internal" href="#discussion">discussion</a> listed at the end of this PEP. </section> <section id="abstract"> <h2><a class="toc-backref" href="#abstract" role="doc-backlink">Abstract</a></h2> This PEP describes an addition to the standard library called Data Classes. Although they use a very different mechanism, Data Classes can be thought of as “mutable namedtuples with defaults”. Because Data Classes use normal class definition syntax, you are free to use inheritance, metaclasses, docstrings, user-defined methods, class factories, and other Python class features. A class decorator is provided which inspects a class definition for variables with type annotations as defined in <a class="pep reference internal" href="../pep-0526/" title="PEP 526 – Syntax for Variable Annotations">PEP 526</a>, “Syntax for Variable Annotations”. In this document, such variables are called fields. Using these fields, the decorator adds generated method definitions to the class to support instance initialization, a repr, comparison methods, and optionally other methods as described in the <a class="reference internal" href="#specification">Specification</a> section. Such a class is called a Data Class, but there’s really nothing special about the class: the decorator adds generated methods to the class and returns the same class it was given. As an example: <div class="highlight-default notranslate"><div class="highlight"><pre>@dataclass class InventoryItem: '''Class for keeping track of an item in inventory.''' name: str unit_price: float quantity_on_hand: int = 0 def total_cost(self) -> float: return self.unit_price * self.quantity_on_hand </pre></div> </div> The <code class="docutils literal notranslate">@dataclass</code> decorator will add the equivalent of these methods to the InventoryItem class: <div class="highlight-default notranslate"><div class="highlight"><pre>def __init__(self, name: str, unit_price: float, quantity_on_hand: int = 0) -> None: self.name = name self.unit_price = unit_price self.quantity_on_hand = quantity_on_hand def __repr__(self): return f'InventoryItem(name={self.name!r}, unit_price={self.unit_price!r}, quantity_on_hand={self.quantity_on_hand!r})' def __eq__(self, other): if other.__class__ is self.__class__: return (self.name, self.unit_price, self.quantity_on_hand) == (other.name, other.unit_price, other.quantity_on_hand) return NotImplemented def __ne__(self, other): if other.__class__ is self.__class__: return (self.name, self.unit_price, self.quantity_on_hand) != (other.name, other.unit_price, other.quantity_on_hand) return NotImplemented def __lt__(self, other): if other.__class__ is self.__class__: return (self.name, self.unit_price, self.quantity_on_hand) < (other.name, other.unit_price, other.quantity_on_hand) return NotImplemented def __le__(self, other): if other.__class__ is self.__class__: return (self.name, self.unit_price, self.quantity_on_hand) <= (other.name, other.unit_price, other.quantity_on_hand) return NotImplemented def __gt__(self, other): if other.__class__ is self.__class__: return (self.name, self.unit_price, self.quantity_on_hand) > (other.name, other.unit_price, other.quantity_on_hand) return NotImplemented def __ge__(self, other): if other.__class__ is self.__class__: return (self.name, self.unit_price, self.quantity_on_hand) >= (other.name, other.unit_price, other.quantity_on_hand) return NotImplemented </pre></div> </div> Data Classes save you from writing and maintaining these methods. </section> <section id="rationale"> <h2><a class="toc-backref" href="#rationale" role="doc-backlink">Rationale</a></h2> There have been numerous attempts to define classes which exist primarily to store values which are accessible by attribute lookup. Some examples include: <ul class="simple"> <li>collections.namedtuple in the standard library.</li> <li>typing.NamedTuple in the standard library.</li> <li>The popular attrs <a class="footnote-reference brackets" href="#id15" id="id1">[1]</a> project.</li> <li>George Sakkis’ recordType recipe <a class="footnote-reference brackets" href="#id16" id="id2">[2]</a>, a mutable data type inspired by collections.namedtuple.</li> <li>Many example online recipes <a class="footnote-reference brackets" href="#id17" id="id3">[3]</a>, packages <a class="footnote-reference brackets" href="#id18" id="id4">[4]</a>, and questions <a class="footnote-reference brackets" href="#id19" id="id5">[5]</a>. David Beazley used a form of data classes as the motivating example in a PyCon 2013 metaclass talk <a class="footnote-reference brackets" href="#id20" id="id6">[6]</a>.</li> </ul> So, why is this PEP needed? With the addition of <a class="pep reference internal" href="../pep-0526/" title="PEP 526 – Syntax for Variable Annotations">PEP 526</a>, Python has a concise way to specify the type of class members. This PEP leverages that syntax to provide a simple, unobtrusive way to describe Data Classes. With two exceptions, the specified attribute type annotation is completely ignored by Data Classes. No base classes or metaclasses are used by Data Classes. Users of these classes are free to use inheritance and metaclasses without any interference from Data Classes. The decorated classes are truly “normal” Python classes. The Data Class decorator should not interfere with any usage of the class. One main design goal of Data Classes is to support static type checkers. The use of <a class="pep reference internal" href="../pep-0526/" title="PEP 526 – Syntax for Variable Annotations">PEP 526</a> syntax is one example of this, but so is the design of the <code class="docutils literal notranslate">fields()</code> function and the <code class="docutils literal notranslate">@dataclass</code> decorator. Due to their very dynamic nature, some of the libraries mentioned above are difficult to use with static type checkers. Data Classes are not, and are not intended to be, a replacement mechanism for all of the above libraries. But being in the standard library will allow many of the simpler use cases to instead leverage Data Classes. Many of the libraries listed have different feature sets, and will of course continue to exist and prosper. Where is it not appropriate to use Data Classes? <ul class="simple"> <li>API compatibility with tuples or dicts is required.</li> <li>Type validation beyond that provided by PEPs 484 and 526 is required, or value validation or conversion is required.</li> </ul> </section> <section id="specification"> <h2><a class="toc-backref" href="#specification" role="doc-backlink">Specification</a></h2> All of the functions described in this PEP will live in a module named <code class="docutils literal notranslate">dataclasses</code>. A function <code class="docutils literal notranslate">dataclass</code> which is typically used as a class decorator is provided to post-process classes and add generated methods, described below. The <code class="docutils literal notranslate">dataclass</code> decorator examines the class to find <code class="docutils literal notranslate">field</code>s. A <code class="docutils literal notranslate">field</code> is defined as any variable identified in <code class="docutils literal notranslate">__annotations__</code>. That is, a variable that has a type annotation. With two exceptions described below, none of the Data Class machinery examines the type specified in the annotation. Note that <code class="docutils literal notranslate">__annotations__</code> is guaranteed to be an ordered mapping, in class declaration order. The order of the fields in all of the generated methods is the order in which they appear in the class. The <code class="docutils literal notranslate">dataclass</code> decorator will add various “dunder” methods to the class, described below. If any of the added methods already exist on the class, a <code class="docutils literal notranslate">TypeError</code> will be raised. The decorator returns the same class that is called on: no new class is created. The <code class="docutils literal notranslate">dataclass</code> decorator is typically used with no parameters and no parentheses. However, it also supports the following logical signature: <div class="highlight-default notranslate"><div class="highlight"><pre>def dataclass(*, init=True, repr=True, eq=True, order=False, unsafe_hash=False, frozen=False) </pre></div> </div> If <code class="docutils literal notranslate">dataclass</code> is used just as a simple decorator with no parameters, it acts as if it has the default values documented in this signature. That is, these three uses of <code class="docutils literal notranslate">@dataclass</code> are equivalent: <div class="highlight-default notranslate"><div class="highlight"><pre>@dataclass class C: ... @dataclass() class C: ... @dataclass(init=True, repr=True, eq=True, order=False, unsafe_hash=False, frozen=False) class C: ... </pre></div> </div> The parameters to <code class="docutils literal notranslate">dataclass</code> are: <ul> <li><code class="docutils literal notranslate">init</code>: If true (the default), a <code class="docutils literal notranslate">__init__</code> method will be generated.</li> <li><code class="docutils literal notranslate">repr</code>: If true (the default), a <code class="docutils literal notranslate">__repr__</code> method will be generated. The generated repr string will have the class name and the name and repr of each field, in the order they are defined in the class. Fields that are marked as being excluded from the repr are not included. For example: <code class="docutils literal notranslate">InventoryItem(name='widget', unit_price=3.0, quantity_on_hand=10)</code>.If the class already defines <code class="docutils literal notranslate">__repr__</code>, this parameter is ignored. </li> <li><code class="docutils literal notranslate">eq</code>: If true (the default), an <code class="docutils literal notranslate">__eq__</code> method will be generated. This method compares the class as if it were a tuple of its fields, in order. Both instances in the comparison must be of the identical type.If the class already defines <code class="docutils literal notranslate">__eq__</code>, this parameter is ignored. </li> <li><code class="docutils literal notranslate">order</code>: If true (the default is False), <code class="docutils literal notranslate">__lt__</code>, <code class="docutils literal notranslate">__le__</code>, <code class="docutils literal notranslate">__gt__</code>, and <code class="docutils literal notranslate">__ge__</code> methods will be generated. These compare the class as if it were a tuple of its fields, in order. Both instances in the comparison must be of the identical type. If <code class="docutils literal notranslate">order</code> is true and <code class="docutils literal notranslate">eq</code> is false, a <code class="docutils literal notranslate">ValueError</code> is raised.If the class already defines any of <code class="docutils literal notranslate">__lt__</code>, <code class="docutils literal notranslate">__le__</code>, <code class="docutils literal notranslate">__gt__</code>, or <code class="docutils literal notranslate">__ge__</code>, then <code class="docutils literal notranslate">ValueError</code> is raised. </li> <li><code class="docutils literal notranslate">unsafe_hash</code>: If <code class="docutils literal notranslate">False</code> (the default), the <code class="docutils literal notranslate">__hash__</code> method is generated according to how <code class="docutils literal notranslate">eq</code> and <code class="docutils literal notranslate">frozen</code> are set.If <code class="docutils literal notranslate">eq</code> and <code class="docutils literal notranslate">frozen</code> are both true, Data Classes will generate a <code class="docutils literal notranslate">__hash__</code> method for you. If <code class="docutils literal notranslate">eq</code> is true and <code class="docutils literal notranslate">frozen</code> is false, <code class="docutils literal notranslate">__hash__</code> will be set to <code class="docutils literal notranslate">None</code>, marking it unhashable (which it is). If <code class="docutils literal notranslate">eq</code> is false, <code class="docutils literal notranslate">__hash__</code> will be left untouched meaning the <code class="docutils literal notranslate">__hash__</code> method of the superclass will be used (if the superclass is <code class="docutils literal notranslate">object</code>, this means it will fall back to id-based hashing). Although not recommended, you can force Data Classes to create a <code class="docutils literal notranslate">__hash__</code> method with <code class="docutils literal notranslate">unsafe_hash=True</code>. This might be the case if your class is logically immutable but can nonetheless be mutated. This is a specialized use case and should be considered carefully. If a class already has an explicitly defined <code class="docutils literal notranslate">__hash__</code> the behavior when adding <code class="docutils literal notranslate">__hash__</code> is modified. An explicitly defined <code class="docutils literal notranslate">__hash__</code> is defined when: <blockquote> <div><ul class="simple"> <li><code class="docutils literal notranslate">__eq__</code> is defined in the class and <code class="docutils literal notranslate">__hash__</code> is defined with any value other than <code class="docutils literal notranslate">None</code>.</li> <li><code class="docutils literal notranslate">__eq__</code> is defined in the class and any non-<code class="docutils literal notranslate">None</code> <code class="docutils literal notranslate">__hash__</code> is defined.</li> <li><code class="docutils literal notranslate">__eq__</code> is not defined on the class, and any <code class="docutils literal notranslate">__hash__</code> is defined.</li> </ul> </div></blockquote> If <code class="docutils literal notranslate">unsafe_hash</code> is true and an explicitly defined <code class="docutils literal notranslate">__hash__</code> is present, then <code class="docutils literal notranslate">ValueError</code> is raised. If <code class="docutils literal notranslate">unsafe_hash</code> is false and an explicitly defined <code class="docutils literal notranslate">__hash__</code> is present, then no <code class="docutils literal notranslate">__hash__</code> is added. See the Python documentation <a class="footnote-reference brackets" href="#id21" id="id7">[7]</a> for more information. </li> <li><code class="docutils literal notranslate">frozen</code>: If true (the default is False), assigning to fields will generate an exception. This emulates read-only frozen instances. If either <code class="docutils literal notranslate">__getattr__</code> or <code class="docutils literal notranslate">__setattr__</code> is defined in the class, then <code class="docutils literal notranslate">ValueError</code> is raised. See the discussion below.</li> </ul> <code class="docutils literal notranslate">field</code>s may optionally specify a default value, using normal Python syntax: <div class="highlight-default notranslate"><div class="highlight"><pre>@dataclass class C: a: int # 'a' has no default value b: int = 0 # assign a default value for 'b' </pre></div> </div> In this example, both <code class="docutils literal notranslate">a</code> and <code class="docutils literal notranslate">b</code> will be included in the added <code class="docutils literal notranslate">__init__</code> method, which will be defined as: <div class="highlight-default notranslate"><div class="highlight"><pre>def __init__(self, a: int, b: int = 0): </pre></div> </div> <code class="docutils literal notranslate">TypeError</code> will be raised if a field without a default value follows a field with a default value. This is true either when this occurs in a single class, or as a result of class inheritance. For common and simple use cases, no other functionality is required. There are, however, some Data Class features that require additional per-field information. To satisfy this need for additional information, you can replace the default field value with a call to the provided <code class="docutils literal notranslate">field()</code> function. The signature of <code class="docutils literal notranslate">field()</code> is: <div class="highlight-default notranslate"><div class="highlight"><pre>def field(*, default=MISSING, default_factory=MISSING, repr=True, hash=None, init=True, compare=True, metadata=None) </pre></div> </div> The <code class="docutils literal notranslate">MISSING</code> value is a sentinel object used to detect if the <code class="docutils literal notranslate">default</code> and <code class="docutils literal notranslate">default_factory</code> parameters are provided. This sentinel is used because <code class="docutils literal notranslate">None</code> is a valid value for <code class="docutils literal notranslate">default</code>. The parameters to <code class="docutils literal notranslate">field()</code> are: <ul> <li><code class="docutils literal notranslate">default</code>: If provided, this will be the default value for this field. This is needed because the <code class="docutils literal notranslate">field</code> call itself replaces the normal position of the default value.</li> <li><code class="docutils literal notranslate">default_factory</code>: If provided, it must be a zero-argument callable that will be called when a default value is needed for this field. Among other purposes, this can be used to specify fields with mutable default values, as discussed below. It is an error to specify both <code class="docutils literal notranslate">default</code> and <code class="docutils literal notranslate">default_factory</code>.</li> <li><code class="docutils literal notranslate">init</code>: If true (the default), this field is included as a parameter to the generated <code class="docutils literal notranslate">__init__</code> method.</li> <li><code class="docutils literal notranslate">repr</code>: If true (the default), this field is included in the string returned by the generated <code class="docutils literal notranslate">__repr__</code> method.</li> <li><code class="docutils literal notranslate">compare</code>: If True (the default), this field is included in the generated equality and comparison methods (<code class="docutils literal notranslate">__eq__</code>, <code class="docutils literal notranslate">__gt__</code>, et al.).</li> <li><code class="docutils literal notranslate">hash</code>: This can be a bool or <code class="docutils literal notranslate">None</code>. If True, this field is included in the generated <code class="docutils literal notranslate">__hash__</code> method. If <code class="docutils literal notranslate">None</code> (the default), use the value of <code class="docutils literal notranslate">compare</code>: this would normally be the expected behavior. A field should be considered in the hash if it’s used for comparisons. Setting this value to anything other than <code class="docutils literal notranslate">None</code> is discouraged.One possible reason to set <code class="docutils literal notranslate">hash=False</code> but <code class="docutils literal notranslate">compare=True</code> would be if a field is expensive to compute a hash value for, that field is needed for equality testing, and there are other fields that contribute to the type’s hash value. Even if a field is excluded from the hash, it will still be used for comparisons. </li> <li><code class="docutils literal notranslate">metadata</code>: This can be a mapping or None. None is treated as an empty dict. This value is wrapped in <code class="docutils literal notranslate">types.MappingProxyType</code> to make it read-only, and exposed on the Field object. It is not used at all by Data Classes, and is provided as a third-party extension mechanism. Multiple third-parties can each have their own key, to use as a namespace in the metadata.</li> </ul> If the default value of a field is specified by a call to <code class="docutils literal notranslate">field()</code>, then the class attribute for this field will be replaced by the specified <code class="docutils literal notranslate">default</code> value. If no <code class="docutils literal notranslate">default</code> is provided, then the class attribute will be deleted. The intent is that after the <code class="docutils literal notranslate">dataclass</code> decorator runs, the class attributes will all contain the default values for the fields, just as if the default value itself were specified. For example, after: <div class="highlight-default notranslate"><div class="highlight"><pre>@dataclass class C: x: int y: int = field(repr=False) z: int = field(repr=False, default=10) t: int = 20 </pre></div> </div> The class attribute <code class="docutils literal notranslate">C.z</code> will be <code class="docutils literal notranslate">10</code>, the class attribute <code class="docutils literal notranslate">C.t</code> will be <code class="docutils literal notranslate">20</code>, and the class attributes <code class="docutils literal notranslate">C.x</code> and <code class="docutils literal notranslate">C.y</code> will not be set. <section id="field-objects"> <h3><a class="toc-backref" href="#field-objects" role="doc-backlink"><code class="docutils literal notranslate">Field</code> objects</a></h3> <code class="docutils literal notranslate">Field</code> objects describe each defined field. These objects are created internally, and are returned by the <code class="docutils literal notranslate">fields()</code> module-level method (see below). Users should never instantiate a <code class="docutils literal notranslate">Field</code> object directly. Its documented attributes are: <ul class="simple"> <li><code class="docutils literal notranslate">name</code>: The name of the field.</li> <li><code class="docutils literal notranslate">type</code>: The type of the field.</li> <li><code class="docutils literal notranslate">default</code>, <code class="docutils literal notranslate">default_factory</code>, <code class="docutils literal notranslate">init</code>, <code class="docutils literal notranslate">repr</code>, <code class="docutils literal notranslate">hash</code>, <code class="docutils literal notranslate">compare</code>, and <code class="docutils literal notranslate">metadata</code> have the identical meaning and values as they do in the <code class="docutils literal notranslate">field()</code> declaration.</li> </ul> Other attributes may exist, but they are private and must not be inspected or relied on. </section> <section id="post-init-processing"> <h3><a class="toc-backref" href="#post-init-processing" role="doc-backlink">post-init processing</a></h3> The generated <code class="docutils literal notranslate">__init__</code> code will call a method named <code class="docutils literal notranslate">__post_init__</code>, if it is defined on the class. It will be called as <code class="docutils literal notranslate">self.__post_init__()</code>. If no <code class="docutils literal notranslate">__init__</code> method is generated, then <code class="docutils literal notranslate">__post_init__</code> will not automatically be called. Among other uses, this allows for initializing field values that depend on one or more other fields. For example: <div class="highlight-default notranslate"><div class="highlight"><pre>@dataclass class C: a: float b: float c: float = field(init=False) def __post_init__(self): self.c = self.a + self.b </pre></div> </div> See the section below on init-only variables for ways to pass parameters to <code class="docutils literal notranslate">__post_init__()</code>. Also see the warning about how <code class="docutils literal notranslate">replace()</code> handles <code class="docutils literal notranslate">init=False</code> fields. </section> <section id="class-variables"> <h3><a class="toc-backref" href="#class-variables" role="doc-backlink">Class variables</a></h3> One place where <code class="docutils literal notranslate">dataclass</code> actually inspects the type of a field is to determine if a field is a class variable as defined in <a class="pep reference internal" href="../pep-0526/" title="PEP 526 – Syntax for Variable Annotations">PEP 526</a>. It does this by checking if the type of the field is <code class="docutils literal notranslate">typing.ClassVar</code>. If a field is a <code class="docutils literal notranslate">ClassVar</code>, it is excluded from consideration as a field and is ignored by the Data Class mechanisms. For more discussion, see <a class="footnote-reference brackets" href="#id22" id="id8">[8]</a>. Such <code class="docutils literal notranslate">ClassVar</code> pseudo-fields are not returned by the module-level <code class="docutils literal notranslate">fields()</code> function. </section> <section id="init-only-variables"> <h3><a class="toc-backref" href="#init-only-variables" role="doc-backlink">Init-only variables</a></h3> The other place where <code class="docutils literal notranslate">dataclass</code> inspects a type annotation is to determine if a field is an init-only variable. It does this by seeing if the type of a field is of type <code class="docutils literal notranslate">dataclasses.InitVar</code>. If a field is an <code class="docutils literal notranslate">InitVar</code>, it is considered a pseudo-field called an init-only field. As it is not a true field, it is not returned by the module-level <code class="docutils literal notranslate">fields()</code> function. Init-only fields are added as parameters to the generated <code class="docutils literal notranslate">__init__</code> method, and are passed to the optional <code class="docutils literal notranslate">__post_init__</code> method. They are not otherwise used by Data Classes. For example, suppose a field will be initialized from a database, if a value is not provided when creating the class: <div class="highlight-default notranslate"><div class="highlight"><pre>@dataclass class C: i: int j: int = None database: InitVar[DatabaseType] = None def __post_init__(self, database): if self.j is None and database is not None: self.j = database.lookup('j') c = C(10, database=my_database) </pre></div> </div> In this case, <code class="docutils literal notranslate">fields()</code> will return <code class="docutils literal notranslate">Field</code> objects for <code class="docutils literal notranslate">i</code> and <code class="docutils literal notranslate">j</code>, but not for <code class="docutils literal notranslate">database</code>. </section> <section id="frozen-instances"> <h3><a class="toc-backref" href="#frozen-instances" role="doc-backlink">Frozen instances</a></h3> It is not possible to create truly immutable Python objects. However, by passing <code class="docutils literal notranslate">frozen=True</code> to the <code class="docutils literal notranslate">@dataclass</code> decorator you can emulate immutability. In that case, Data Classes will add <code class="docutils literal notranslate">__setattr__</code> and <code class="docutils literal notranslate">__delattr__</code> methods to the class. These methods will raise a <code class="docutils literal notranslate">FrozenInstanceError</code> when invoked. There is a tiny performance penalty when using <code class="docutils literal notranslate">frozen=True</code>: <code class="docutils literal notranslate">__init__</code> cannot use simple assignment to initialize fields, and must use <code class="docutils literal notranslate">object.__setattr__</code>. </section> <section id="inheritance"> <h3><a class="toc-backref" href="#inheritance" role="doc-backlink">Inheritance</a></h3> When the Data Class is being created by the <code class="docutils literal notranslate">@dataclass</code> decorator, it looks through all of the class’s base classes in reverse MRO (that is, starting at <code class="docutils literal notranslate">object</code>) and, for each Data Class that it finds, adds the fields from that base class to an ordered mapping of fields. After all of the base class fields are added, it adds its own fields to the ordered mapping. All of the generated methods will use this combined, calculated ordered mapping of fields. Because the fields are in insertion order, derived classes override base classes. An example: <div class="highlight-default notranslate"><div class="highlight"><pre>@dataclass class Base: x: Any = 15.0 y: int = 0 @dataclass class C(Base): z: int = 10 x: int = 15 </pre></div> </div> The final list of fields is, in order, <code class="docutils literal notranslate">x</code>, <code class="docutils literal notranslate">y</code>, <code class="docutils literal notranslate">z</code>. The final type of <code class="docutils literal notranslate">x</code> is <code class="docutils literal notranslate">int</code>, as specified in class <code class="docutils literal notranslate">C</code>. The generated <code class="docutils literal notranslate">__init__</code> method for <code class="docutils literal notranslate">C</code> will look like: <div class="highlight-default notranslate"><div class="highlight"><pre>def __init__(self, x: int = 15, y: int = 0, z: int = 10): </pre></div> </div> </section> <section id="default-factory-functions"> <h3><a class="toc-backref" href="#default-factory-functions" role="doc-backlink">Default factory functions</a></h3> If a field specifies a <code class="docutils literal notranslate">default_factory</code>, it is called with zero arguments when a default value for the field is needed. For example, to create a new instance of a list, use: <div class="highlight-default notranslate"><div class="highlight"><pre>l: list = field(default_factory=list) </pre></div> </div> If a field is excluded from <code class="docutils literal notranslate">__init__</code> (using <code class="docutils literal notranslate">init=False</code>) and the field also specifies <code class="docutils literal notranslate">default_factory</code>, then the default factory function will always be called from the generated <code class="docutils literal notranslate">__init__</code> function. This happens because there is no other way to give the field an initial value. </section> <section id="mutable-default-values"> <h3><a class="toc-backref" href="#mutable-default-values" role="doc-backlink">Mutable default values</a></h3> Python stores default member variable values in class attributes. Consider this example, not using Data Classes: <div class="highlight-default notranslate"><div class="highlight"><pre>class C: x = [] def add(self, element): self.x += element o1 = C() o2 = C() o1.add(1) o2.add(2) assert o1.x == [1, 2] assert o1.x is o2.x </pre></div> </div> Note that the two instances of class <code class="docutils literal notranslate">C</code> share the same class variable <code class="docutils literal notranslate">x</code>, as expected. Using Data Classes, if this code was valid: <div class="highlight-default notranslate"><div class="highlight"><pre>@dataclass class D: x: List = [] def add(self, element): self.x += element </pre></div> </div> it would generate code similar to: <div class="highlight-default notranslate"><div class="highlight"><pre>class D: x = [] def __init__(self, x=x): self.x = x def add(self, element): self.x += element assert D().x is D().x </pre></div> </div> This has the same issue as the original example using class <code class="docutils literal notranslate">C</code>. That is, two instances of class <code class="docutils literal notranslate">D</code> that do not specify a value for <code class="docutils literal notranslate">x</code> when creating a class instance will share the same copy of <code class="docutils literal notranslate">x</code>. Because Data Classes just use normal Python class creation they also share this problem. There is no general way for Data Classes to detect this condition. Instead, Data Classes will raise a <code class="docutils literal notranslate">TypeError</code> if it detects a default parameter of type <code class="docutils literal notranslate">list</code>, <code class="docutils literal notranslate">dict</code>, or <code class="docutils literal notranslate">set</code>. This is a partial solution, but it does protect against many common errors. See <a class="reference internal" href="#automatically-support-mutable-default-values">Automatically support mutable default values</a> in the Rejected Ideas section for more details. Using default factory functions is a way to create new instances of mutable types as default values for fields: <div class="highlight-default notranslate"><div class="highlight"><pre>@dataclass class D: x: list = field(default_factory=list) assert D().x is not D().x </pre></div> </div> </section> <section id="module-level-helper-functions"> <h3><a class="toc-backref" href="#module-level-helper-functions" role="doc-backlink">Module level helper functions</a></h3> <ul> <li><code class="docutils literal notranslate">fields(class_or_instance)</code>: Returns a tuple of <code class="docutils literal notranslate">Field</code> objects that define the fields for this Data Class. Accepts either a Data Class, or an instance of a Data Class. Raises <code class="docutils literal notranslate">ValueError</code> if not passed a Data Class or instance of one. Does not return pseudo-fields which are <code class="docutils literal notranslate">ClassVar</code> or <code class="docutils literal notranslate">InitVar</code>.</li> <li><code class="docutils literal notranslate">asdict(instance, *, dict_factory=dict)</code>: Converts the Data Class <code class="docutils literal notranslate">instance</code> to a dict (by using the factory function <code class="docutils literal notranslate">dict_factory</code>). Each Data Class is converted to a dict of its fields, as name:value pairs. Data Classes, dicts, lists, and tuples are recursed into. For example:<div class="highlight-default notranslate"><div class="highlight"><pre>@dataclass class Point: x: int y: int @dataclass class C: l: List[Point] p = Point(10, 20) assert asdict(p) == {'x': 10, 'y': 20} c = C([Point(0, 0), Point(10, 4)]) assert asdict(c) == {'l': [{'x': 0, 'y': 0}, {'x': 10, 'y': 4}]} </pre></div> </div> Raises <code class="docutils literal notranslate">TypeError</code> if <code class="docutils literal notranslate">instance</code> is not a Data Class instance. </li> <li><code class="docutils literal notranslate">astuple(*, tuple_factory=tuple)</code>: Converts the Data Class <code class="docutils literal notranslate">instance</code> to a tuple (by using the factory function <code class="docutils literal notranslate">tuple_factory</code>). Each Data Class is converted to a tuple of its field values. Data Classes, dicts, lists, and tuples are recursed into.Continuing from the previous example: <div class="highlight-default notranslate"><div class="highlight"><pre>assert astuple(p) == (10, 20) assert astuple(c) == ([(0, 0), (10, 4)],) </pre></div> </div> Raises <code class="docutils literal notranslate">TypeError</code> if <code class="docutils literal notranslate">instance</code> is not a Data Class instance. </li> <li><code class="docutils literal notranslate">make_dataclass(cls_name, fields, *, bases=(), namespace=None)</code>: Creates a new Data Class with name <code class="docutils literal notranslate">cls_name</code>, fields as defined in <code class="docutils literal notranslate">fields</code>, base classes as given in <code class="docutils literal notranslate">bases</code>, and initialized with a namespace as given in <code class="docutils literal notranslate">namespace</code>. <code class="docutils literal notranslate">fields</code> is an iterable whose elements are either <code class="docutils literal notranslate">name</code>, <code class="docutils literal notranslate">(name, type)</code>, or <code class="docutils literal notranslate">(name, type, Field)</code>. If just <code class="docutils literal notranslate">name</code> is supplied, <code class="docutils literal notranslate">typing.Any</code> is used for <code class="docutils literal notranslate">type</code>. This function is not strictly required, because any Python mechanism for creating a new class with <code class="docutils literal notranslate">__annotations__</code> can then apply the <code class="docutils literal notranslate">dataclass</code> function to convert that class to a Data Class. This function is provided as a convenience. For example:<div class="highlight-default notranslate"><div class="highlight"><pre>C = make_dataclass('C', [('x', int), 'y', ('z', int, field(default=5))], namespace={'add_one': lambda self: self.x + 1}) </pre></div> </div> Is equivalent to: <div class="highlight-default notranslate"><div class="highlight"><pre>@dataclass class C: x: int y: 'typing.Any' z: int = 5 def add_one(self): return self.x + 1 </pre></div> </div> </li> <li><code class="docutils literal notranslate">replace(instance, **changes)</code>: Creates a new object of the same type of <code class="docutils literal notranslate">instance</code>, replacing fields with values from <code class="docutils literal notranslate">changes</code>. If <code class="docutils literal notranslate">instance</code> is not a Data Class, raises <code class="docutils literal notranslate">TypeError</code>. If values in <code class="docutils literal notranslate">changes</code> do not specify fields, raises <code class="docutils literal notranslate">TypeError</code>.The newly returned object is created by calling the <code class="docutils literal notranslate">__init__</code> method of the Data Class. This ensures that <code class="docutils literal notranslate">__post_init__</code>, if present, is also called. Init-only variables without default values, if any exist, must be specified on the call to <code class="docutils literal notranslate">replace</code> so that they can be passed to <code class="docutils literal notranslate">__init__</code> and <code class="docutils literal notranslate">__post_init__</code>. It is an error for <code class="docutils literal notranslate">changes</code> to contain any fields that are defined as having <code class="docutils literal notranslate">init=False</code>. A <code class="docutils literal notranslate">ValueError</code> will be raised in this case. Be forewarned about how <code class="docutils literal notranslate">init=False</code> fields work during a call to <code class="docutils literal notranslate">replace()</code>. They are not copied from the source object, but rather are initialized in <code class="docutils literal notranslate">__post_init__()</code>, if they’re initialized at all. It is expected that <code class="docutils literal notranslate">init=False</code> fields will be rarely and judiciously used. If they are used, it might be wise to have alternate class constructors, or perhaps a custom <code class="docutils literal notranslate">replace()</code> (or similarly named) method which handles instance copying. </li> <li><code class="docutils literal notranslate">is_dataclass(class_or_instance)</code>: Returns True if its parameter is a dataclass or an instance of one, otherwise returns False.If you need to know if a class is an instance of a dataclass (and not a dataclass itself), then add a further check for <code class="docutils literal notranslate">not isinstance(obj, type)</code>: <div class="highlight-default notranslate"><div class="highlight"><pre>def is_dataclass_instance(obj): return is_dataclass(obj) and not isinstance(obj, type) </pre></div> </div> </li> </ul> </section> </section> <section id="discussion"> <h2><a class="toc-backref" href="#discussion" role="doc-backlink">Discussion</a></h2> <section id="python-ideas-discussion"> <h3><a class="toc-backref" href="#python-ideas-discussion" role="doc-backlink">python-ideas discussion</a></h3> This discussion started on python-ideas <a class="footnote-reference brackets" href="#id23" id="id9">[9]</a> and was moved to a GitHub repo <a class="footnote-reference brackets" href="#id24" id="id10">[10]</a> for further discussion. As part of this discussion, we made the decision to use <a class="pep reference internal" href="../pep-0526/" title="PEP 526 – Syntax for Variable Annotations">PEP 526</a> syntax to drive the discovery of fields. </section> <section id="support-for-automatically-setting-slots"> <h3><a class="toc-backref" href="#support-for-automatically-setting-slots" role="doc-backlink">Support for automatically setting <code class="docutils literal notranslate">__slots__</code>?</a></h3> At least for the initial release, <code class="docutils literal notranslate">__slots__</code> will not be supported. <code class="docutils literal notranslate">__slots__</code> needs to be added at class creation time. The Data Class decorator is called after the class is created, so in order to add <code class="docutils literal notranslate">__slots__</code> the decorator would have to create a new class, set <code class="docutils literal notranslate">__slots__</code>, and return it. Because this behavior is somewhat surprising, the initial version of Data Classes will not support automatically setting <code class="docutils literal notranslate">__slots__</code>. There are a number of workarounds: <ul class="simple"> <li>Manually add <code class="docutils literal notranslate">__slots__</code> in the class definition.</li> <li>Write a function (which could be used as a decorator) that inspects the class using <code class="docutils literal notranslate">fields()</code> and creates a new class with <code class="docutils literal notranslate">__slots__</code> set.</li> </ul> For more discussion, see <a class="footnote-reference brackets" href="#id25" id="id11">[11]</a>. </section> <section id="why-not-just-use-namedtuple"> <h3><a class="toc-backref" href="#why-not-just-use-namedtuple" role="doc-backlink">Why not just use namedtuple?</a></h3> <ul> <li>Any namedtuple can be accidentally compared to any other with the same number of fields. For example: <code class="docutils literal notranslate">Point3D(2017, 6, 2) == Date(2017, 6, 2)</code>. With Data Classes, this would return False.</li> <li>A namedtuple can be accidentally compared to a tuple. For example, <code class="docutils literal notranslate">Point2D(1, 10) == (1, 10)</code>. With Data Classes, this would return False.</li> <li>Instances are always iterable, which can make it difficult to add fields. If a library defines:<div class="highlight-default notranslate"><div class="highlight"><pre>Time = namedtuple('Time', ['hour', 'minute']) def get_time(): return Time(12, 0) </pre></div> </div> Then if a user uses this code as: <div class="highlight-default notranslate"><div class="highlight"><pre>hour, minute = get_time() </pre></div> </div> then it would not be possible to add a <code class="docutils literal notranslate">second</code> field to <code class="docutils literal notranslate">Time</code> without breaking the user’s code. </li> <li>No option for mutable instances.</li> <li>Cannot specify default values.</li> <li>Cannot control which fields are used for <code class="docutils literal notranslate">__init__</code>, <code class="docutils literal notranslate">__repr__</code>, etc.</li> <li>Cannot support combining fields by inheritance.</li> </ul> </section> <section id="why-not-just-use-typing-namedtuple"> <h3><a class="toc-backref" href="#why-not-just-use-typing-namedtuple" role="doc-backlink">Why not just use typing.NamedTuple?</a></h3> For classes with statically defined fields, it does support similar syntax to Data Classes, using type annotations. This produces a namedtuple, so it shares <code class="docutils literal notranslate">namedtuple</code>s benefits and some of its downsides. Data Classes, unlike <code class="docutils literal notranslate">typing.NamedTuple</code>, support combining fields via inheritance. </section> <section id="why-not-just-use-attrs"> <h3><a class="toc-backref" href="#why-not-just-use-attrs" role="doc-backlink">Why not just use attrs?</a></h3> <ul class="simple"> <li>attrs moves faster than could be accommodated if it were moved in to the standard library.</li> <li>attrs supports additional features not being proposed here: validators, converters, metadata, etc. Data Classes makes a tradeoff to achieve simplicity by not implementing these features.</li> </ul> For more discussion, see <a class="footnote-reference brackets" href="#id26" id="id12">[12]</a>. </section> <section id="post-init-parameters"> <h3><a class="toc-backref" href="#post-init-parameters" role="doc-backlink">post-init parameters</a></h3> In an earlier version of this PEP before <code class="docutils literal notranslate">InitVar</code> was added, the post-init function <code class="docutils literal notranslate">__post_init__</code> never took any parameters. The normal way of doing parameterized initialization (and not just with Data Classes) is to provide an alternate classmethod constructor. For example: <div class="highlight-default notranslate"><div class="highlight"><pre>@dataclass class C: x: int @classmethod def from_file(cls, filename): with open(filename) as fl: file_value = int(fl.read()) return C(file_value) c = C.from_file('file.txt') </pre></div> </div> Because the <code class="docutils literal notranslate">__post_init__</code> function is the last thing called in the generated <code class="docutils literal notranslate">__init__</code>, having a classmethod constructor (which can also execute code immediately after constructing the object) is functionally equivalent to being able to pass parameters to a <code class="docutils literal notranslate">__post_init__</code> function. With <code class="docutils literal notranslate">InitVar</code>s, <code class="docutils literal notranslate">__post_init__</code> functions can now take parameters. They are passed first to <code class="docutils literal notranslate">__init__</code> which passes them to <code class="docutils literal notranslate">__post_init__</code> where user code can use them as needed. The only real difference between alternate classmethod constructors and <code class="docutils literal notranslate">InitVar</code> pseudo-fields is in regards to required non-field parameters during object creation. With <code class="docutils literal notranslate">InitVar</code>s, using <code class="docutils literal notranslate">__init__</code> and the module-level <code class="docutils literal notranslate">replace()</code> function <code class="docutils literal notranslate">InitVar</code>s must always be specified. Consider the case where a <code class="docutils literal notranslate">context</code> object is needed to create an instance, but isn’t stored as a field. With alternate classmethod constructors the <code class="docutils literal notranslate">context</code> parameter is always optional, because you could still create the object by going through <code class="docutils literal notranslate">__init__</code> (unless you suppress its creation). Which approach is more appropriate will be application-specific, but both approaches are supported. Another reason for using <code class="docutils literal notranslate">InitVar</code> fields is that the class author can control the order of <code class="docutils literal notranslate">__init__</code> parameters. This is especially important with regular fields and <code class="docutils literal notranslate">InitVar</code> fields that have default values, as all fields with defaults must come after all fields without defaults. A previous design had all init-only fields coming after regular fields. This meant that if any field had a default value, then all init-only fields would have to have defaults values, too. </section> <section id="asdict-and-astuple-function-names"> <h3><a class="toc-backref" href="#asdict-and-astuple-function-names" role="doc-backlink">asdict and astuple function names</a></h3> The names of the module-level helper functions <code class="docutils literal notranslate">asdict()</code> and <code class="docutils literal notranslate">astuple()</code> are arguably not <a class="pep reference internal" href="../pep-0008/" title="PEP 8 – Style Guide for Python Code">PEP 8</a> compliant, and should be <code class="docutils literal notranslate">as_dict()</code> and <code class="docutils literal notranslate">as_tuple()</code>, respectively. However, after discussion <a class="footnote-reference brackets" href="#id27" id="id13">[13]</a> it was decided to keep consistency with <code class="docutils literal notranslate">namedtuple._asdict()</code> and <code class="docutils literal notranslate">attr.asdict()</code>. </section> </section> <section id="rejected-ideas"> <h2><a class="toc-backref" href="#rejected-ideas" role="doc-backlink">Rejected ideas</a></h2> <section id="copying-init-false-fields-after-new-object-creation-in-replace"> <h3><a class="toc-backref" href="#copying-init-false-fields-after-new-object-creation-in-replace" role="doc-backlink">Copying <code class="docutils literal notranslate">init=False</code> fields after new object creation in replace()</a></h3> Fields that are <code class="docutils literal notranslate">init=False</code> are by definition not passed to <code class="docutils literal notranslate">__init__</code>, but instead are initialized with a default value, or by calling a default factory function in <code class="docutils literal notranslate">__init__</code>, or by code in <code class="docutils literal notranslate">__post_init__</code>. A previous version of this PEP specified that <code class="docutils literal notranslate">init=False</code> fields would be copied from the source object to the newly created object after <code class="docutils literal notranslate">__init__</code> returned, but that was deemed to be inconsistent with using <code class="docutils literal notranslate">__init__</code> and <code class="docutils literal notranslate">__post_init__</code> to initialize the new object. For example, consider this case: <div class="highlight-default notranslate"><div class="highlight"><pre>@dataclass class Square: length: float area: float = field(init=False, default=0.0) def __post_init__(self): self.area = self.length * self.length s1 = Square(1.0) s2 = replace(s1, length=2.0) </pre></div> </div> If <code class="docutils literal notranslate">init=False</code> fields were copied from the source to the destination object after <code class="docutils literal notranslate">__post_init__</code> is run, then s2 would end up begin <code class="docutils literal notranslate">Square(length=2.0, area=1.0)</code>, instead of the correct <code class="docutils literal notranslate">Square(length=2.0, area=4.0)</code>. </section> <section id="automatically-support-mutable-default-values"> <h3><a class="toc-backref" href="#automatically-support-mutable-default-values" role="doc-backlink">Automatically support mutable default values</a></h3> One proposal was to automatically copy defaults, so that if a literal list <code class="docutils literal notranslate">[]</code> was a default value, each instance would get a new list. There were undesirable side effects of this decision, so the final decision is to disallow the 3 known built-in mutable types: list, dict, and set. For a complete discussion of this and other options, see <a class="footnote-reference brackets" href="#id28" id="id14">[14]</a>. </section> </section> <section id="examples"> <h2><a class="toc-backref" href="#examples" role="doc-backlink">Examples</a></h2> <section id="custom-init-method"> <h3><a class="toc-backref" href="#custom-init-method" role="doc-backlink">Custom __init__ method</a></h3> Sometimes the generated <code class="docutils literal notranslate">__init__</code> method does not suffice. For example, suppose you wanted to have an object to store <code class="docutils literal notranslate">*args</code> and <code class="docutils literal notranslate">**kwargs</code>: <div class="highlight-default notranslate"><div class="highlight"><pre>@dataclass(init=False) class ArgHolder: args: List[Any] kwargs: Mapping[Any, Any] def __init__(self, *args, **kwargs): self.args = args self.kwargs = kwargs a = ArgHolder(1, 2, three=3) </pre></div> </div> </section> <section id="a-complicated-example"> <h3><a class="toc-backref" href="#a-complicated-example" role="doc-backlink">A complicated example</a></h3> This code exists in a closed source project: <div class="highlight-default notranslate"><div class="highlight"><pre>class Application: def __init__(self, name, requirements, constraints=None, path='', executable_links=None, executables_dir=()): self.name = name self.requirements = requirements self.constraints = {} if constraints is None else constraints self.path = path self.executable_links = [] if executable_links is None else executable_links self.executables_dir = executables_dir self.additional_items = [] def __repr__(self): return f'Application({self.name!r},{self.requirements!r},{self.constraints!r},{self.path!r},{self.executable_links!r},{self.executables_dir!r},{self.additional_items!r})' </pre></div> </div> This can be replaced by: <div class="highlight-default notranslate"><div class="highlight"><pre>@dataclass class Application: name: str requirements: List[Requirement] constraints: Dict[str, str] = field(default_factory=dict) path: str = '' executable_links: List[str] = field(default_factory=list) executable_dir: Tuple[str] = () additional_items: List[str] = field(init=False, default_factory=list) </pre></div> </div> The Data Class version is more declarative, has less code, supports <code class="docutils literal notranslate">typing</code>, and includes the other generated functions. </section> </section> <section id="acknowledgements"> <h2><a class="toc-backref" href="#acknowledgements" role="doc-backlink">Acknowledgements</a></h2> The following people provided invaluable input during the development of this PEP and code: Ivan Levkivskyi, Guido van Rossum, Hynek Schlawack, Raymond Hettinger, and Lisa Roach. I thank them for their time and expertise. A special mention must be made about the <code class="docutils literal notranslate">attrs</code> project. It was a true inspiration for this PEP, and I respect the design decisions they made. </section> <section id="references"> <h2><a class="toc-backref" href="#references" role="doc-backlink">References</a></h2> <aside class="footnote-list brackets"> <aside class="footnote brackets" id="id15" role="doc-footnote"> <dt class="label" id="id15">[<a href="#id1">1</a>]</dt> <dd>attrs project on github (<a class="reference external" href="https://github.com/python-attrs/attrs">https://github.com/python-attrs/attrs</a>)</aside> <aside class="footnote brackets" id="id16" role="doc-footnote"> <dt class="label" id="id16">[<a href="#id2">2</a>]</dt> <dd>George Sakkis’ recordType recipe (<a class="reference external" href="http://code.activestate.com/recipes/576555-records/">http://code.activestate.com/recipes/576555-records/</a>)</aside> <aside class="footnote brackets" id="id17" role="doc-footnote"> <dt class="label" id="id17">[<a href="#id3">3</a>]</dt> <dd>DictDotLookup recipe (<a class="reference external" href="http://code.activestate.com/recipes/576586-dot-style-nested-lookups-over-dictionary-based-dat/">http://code.activestate.com/recipes/576586-dot-style-nested-lookups-over-dictionary-based-dat/</a>)</aside> <aside class="footnote brackets" id="id18" role="doc-footnote"> <dt class="label" id="id18">[<a href="#id4">4</a>]</dt> <dd>attrdict package (<a class="reference external" href="https://pypi.python.org/pypi/attrdict">https://pypi.python.org/pypi/attrdict</a>)</aside> <aside class="footnote brackets" id="id19" role="doc-footnote"> <dt class="label" id="id19">[<a href="#id5">5</a>]</dt> <dd>StackOverflow question about data container classes (<a class="reference external" href="https://stackoverflow.com/questions/3357581/using-python-class-as-a-data-container">https://stackoverflow.com/questions/3357581/using-python-class-as-a-data-container</a>)</aside> <aside class="footnote brackets" id="id20" role="doc-footnote"> <dt class="label" id="id20">[<a href="#id6">6</a>]</dt> <dd>David Beazley metaclass talk featuring data classes (<a class="reference external" href="https://www.youtube.com/watch?v=sPiWg5jSoZI">https://www.youtube.com/watch?v=sPiWg5jSoZI</a>)</aside> <aside class="footnote brackets" id="id21" role="doc-footnote"> <dt class="label" id="id21">[<a href="#id7">7</a>]</dt> <dd>Python documentation for __hash__ (<a class="reference external" href="https://docs.python.org/3/reference/datamodel.html#object.__hash__">https://docs.python.org/3/reference/datamodel.html#object.__hash__</a>)</aside> <aside class="footnote brackets" id="id22" role="doc-footnote"> <dt class="label" id="id22">[<a href="#id8">8</a>]</dt> <dd><a class="pep reference internal" href="../pep-0526/#class-and-instance-variable-annotations" title="PEP 526 – Syntax for Variable Annotations § Class and instance variable annotations">ClassVar discussion in PEP 526</a></aside> <aside class="footnote brackets" id="id23" role="doc-footnote"> <dt class="label" id="id23">[<a href="#id9">9</a>]</dt> <dd>Start of python-ideas discussion (<a class="reference external" href="https://mail.python.org/pipermail/python-ideas/2017-May/045618.html">https://mail.python.org/pipermail/python-ideas/2017-May/045618.html</a>)</aside> <aside class="footnote brackets" id="id24" role="doc-footnote"> <dt class="label" id="id24">[<a href="#id10">10</a>]</dt> <dd>GitHub repo where discussions and initial development took place (<a class="reference external" href="https://github.com/ericvsmith/dataclasses">https://github.com/ericvsmith/dataclasses</a>)</aside> <aside class="footnote brackets" id="id25" role="doc-footnote"> <dt class="label" id="id25">[<a href="#id11">11</a>]</dt> <dd>Support __slots__? (<a class="reference external" href="https://github.com/ericvsmith/dataclasses/issues/28">https://github.com/ericvsmith/dataclasses/issues/28</a>)</aside> <aside class="footnote brackets" id="id26" role="doc-footnote"> <dt class="label" id="id26">[<a href="#id12">12</a>]</dt> <dd>why not just attrs? (<a class="reference external" href="https://github.com/ericvsmith/dataclasses/issues/19">https://github.com/ericvsmith/dataclasses/issues/19</a>)</aside> <aside class="footnote brackets" id="id27" role="doc-footnote"> <dt class="label" id="id27">[<a href="#id13">13</a>]</dt> <dd><a class="pep reference internal" href="../pep-0008/" title="PEP 8 – Style Guide for Python Code">PEP 8</a> names for asdict and astuple (<a class="reference external" href="https://github.com/ericvsmith/dataclasses/issues/110">https://github.com/ericvsmith/dataclasses/issues/110</a>)</aside> <aside class="footnote brackets" id="id28" role="doc-footnote"> <dt class="label" id="id28">[<a href="#id14">14</a>]</dt> <dd>Copying mutable defaults (<a class="reference external" href="https://github.com/ericvsmith/dataclasses/issues/3">https://github.com/ericvsmith/dataclasses/issues/3</a>)</aside> </aside> </section> <section id="copyright"> <h2><a class="toc-backref" href="#copyright" role="doc-backlink">Copyright</a></h2> This document has been placed in the public domain. </section> </section> <hr class="docutils" /> Source: <a class="reference external" href="https://github.com/python/peps/blob/main/peps/pep-0557.rst">https://github.com/python/peps/blob/main/peps/pep-0557.rst</a> Last modified: <a class="reference external" href="https://github.com/python/peps/commits/main/peps/pep-0557.rst">2025-02-01 08:55:40 GMT</a> </article> <nav id="pep-sidebar"> <h2>Contents</h2> <ul> <li><a class="reference internal" href="#notice-for-reviewers">Notice for Reviewers</a></li> <li><a class="reference internal" href="#abstract">Abstract</a></li> <li><a class="reference internal" href="#rationale">Rationale</a></li> <li><a class="reference internal" href="#specification">Specification</a><ul> <li><a class="reference internal" href="#field-objects"><code class="docutils literal notranslate">Field</code> objects</a></li> <li><a class="reference internal" href="#post-init-processing">post-init processing</a></li> <li><a class="reference internal" href="#class-variables">Class variables</a></li> <li><a class="reference internal" href="#init-only-variables">Init-only variables</a></li> <li><a class="reference internal" href="#frozen-instances">Frozen instances</a></li> <li><a class="reference internal" href="#inheritance">Inheritance</a></li> <li><a class="reference internal" href="#default-factory-functions">Default factory functions</a></li> <li><a class="reference internal" href="#mutable-default-values">Mutable default values</a></li> <li><a class="reference internal" href="#module-level-helper-functions">Module level helper functions</a></li> </ul> </li> <li><a class="reference internal" href="#discussion">Discussion</a><ul> <li><a class="reference internal" href="#python-ideas-discussion">python-ideas discussion</a></li> <li><a class="reference internal" href="#support-for-automatically-setting-slots">Support for automatically setting <code class="docutils literal notranslate">__slots__</code>?</a></li> <li><a class="reference internal" href="#why-not-just-use-namedtuple">Why not just use namedtuple?</a></li> <li><a class="reference internal" href="#why-not-just-use-typing-namedtuple">Why not just use typing.NamedTuple?</a></li> <li><a class="reference internal" href="#why-not-just-use-attrs">Why not just use attrs?</a></li> <li><a class="reference internal" href="#post-init-parameters">post-init parameters</a></li> <li><a class="reference internal" href="#asdict-and-astuple-function-names">asdict and astuple function names</a></li> </ul> </li> <li><a class="reference internal" href="#rejected-ideas">Rejected ideas</a><ul> <li><a class="reference internal" href="#copying-init-false-fields-after-new-object-creation-in-replace">Copying <code class="docutils literal notranslate">init=False</code> fields after new object creation in replace()</a></li> <li><a class="reference internal" href="#automatically-support-mutable-default-values">Automatically support mutable default values</a></li> </ul> </li> <li><a class="reference internal" href="#examples">Examples</a><ul> <li><a class="reference internal" href="#custom-init-method">Custom __init__ method</a></li> <li><a class="reference internal" href="#a-complicated-example">A complicated example</a></li> </ul> </li> <li><a class="reference internal" href="#acknowledgements">Acknowledgements</a></li> <li><a class="reference internal" href="#references">References</a></li> <li><a class="reference internal" href="#copyright">Copyright</a></li> </ul> <a id="source" href="https://github.com/python/peps/blob/main/peps/pep-0557.rst">Page Source (GitHub)</a> </nav> </section> <script src="../_static/colour_scheme.js"></script> <script src="../_static/wrap_tables.js"></script> <script src="../_static/sticky_banner.js"></script> </body> </html>